Gartner predicts that by 2025, generative AI in healthcare will be producing 10% of all data (currently it is less than 1%) and 20% of all test data for consumer-facing use cases. By 2027, Gartner says 30% of manufacturers will use generative AI in healthcare to enhance their product development effectiveness. Where is generative AI being applied in healthcare and life science settings, and what are some critical factors for implementing it?
Microsoft backed OpenAI released its generative large language model (LLM) ChatGPT in December 2022 and followed that up with the release of GPT-4 in March of 2023. In mid-April 2023, Google announced the release of a medical LLM called Med-Palm2. Their initial release, Med-Palm, had been the first AI system to exceed the 60% pass mark on the U.S. Medical Licensing Exams with a 67.5% score. Med-Palm2 now exceeds 85% accuracy on that same test. This matches very experienced doctor’s ability to provide high quality and authoritative answers to medical questions.
One of the largest deployments of generative Ai in healthcare is in the University of Kansas Health System where more than 1500 doctors across the system’s 140 locations are using Abridge technology that listens in to patient-doctor visits and summarize the most important parts of the conversation. The University of Pittsburgh Medical Center plans to roll out this technology to thousands of their physicians to streamline medical note-taking.
Global professional services company, Avanade, offers five other ways generative AI is impacting healthcare and life sciences.
- It can help diagnose disease and disorders by generating high resolution medical images and synthetic patient data to train machine learning models to recognize and diagnose disease.
- It will personalize medicine by processing and analyzing vast amounts of data that includes genomics, conditions and drug efficacy for optimal, individual treatment plans.
- Generative AI can generate large quantities of electronic chemical representations allowing for chemical structures to be built and tested accelerating the drug discovery process.
- It can generate simulations of medical procedures to train medical robots for accuracy and efficiency reducing the risk of complications.
- It can more effectively predict maintenance needs on medical devices by harvesting the data and analyzing it sooner and quicker.
Avanade goes on to say that getting started in generativeAi in healthcare requires managing the trust and governance portion of the technology first. A “human copilot is essential” and this can be done through appointing responsible, communicative experts in particular areas as AI ambassadors within your firm and selecting use cases where the introduction and integration of OpenAI can be done in a modular and scalable way.